Generalized anxiety disorder (GAD) is a common and debilitating psychiatric condition. The defining feature of this disorder is worrying, which is a future-oriented and maladaptive cognitive process. Cognitive behavioral therapy (CBT) is generally effective for treating GAD, especially when combined with mindfulness approaches. However, CBT is not widely disseminated. In contrast, yoga is widely practiced to enhance mindfulness, reduce stress, and alleviate anxiety. However, little is known about the efficacy of yoga for GAD and how these strategies compare to CBT. In the current application, we propose a five-year multi-site study to examine the comparative efficacy of yoga, CBT, and stress education, a previously employed control condition, for patients with GAD. We propose to randomize 186 patients with GAD to 12-weekly yoga, CBT, or stress education sessions. Independent clinical assessments will occur before the 12-session intervention phase, at mid- treatment, after the intervention, and at 6-month follow-up. We predict greater GAD response and symptom reduction in individuals randomized to receive yoga and those who receive CBT than individuals who receive stress education at post-treatment. Moreover, we predict that the improvement in GAD symptoms in individuals receiving yoga will be comparable to those receiving CBT at post-test. Our secondary aim is to examine the long-term treatment efficacy of yoga on GAD symptoms. We hypothesize that yoga and CBT are equally effective at the 6-month follow-up. Our tertiary aim is to examine and compare the mechanism of treatment changes in yoga and CBT. We predict that treatment changes during yoga are mediated via changes in mindfulness and changes in sympathetic activation, whereas changes in CBT are mediated via changes in maladaptive cognitions. The results of this proposed study have important clinical implications for the practice of yoga and will answer important theoretical questions about the mechanism of treatment change.